SimMIM with MONAI for Self-Supervised Learning on 3D Medical Images

This repository implements SimMIM (Simple Masked Image Modeling) using MONAI, tailored for self-supervised representation learning on 3D medical imaging data.

⚠️ This is a research/experimental notebook primarily run in Google Colab.

---

🧠 What is SimMIM?

SimMIM is a self-supervised learning method that trains a vision transformer (ViT) to reconstruct masked patches of an image. This project adapts that approach for 3D volumetric medical images using MONAI and PyTorch.

---

📁 Files

• data_utils_colab_final_backup.ipynb: Main Colab notebook with SimMIM implementation using MONAI.

---

🚀 Getting Started

To run this notebook:

1. Open it in Google Colab.
2. Make sure to enable GPU runtime.
3. Install MONAI and dependencies:

   pip install monai[all] torch torchvision